This project focuses on the development and implementation of a vision-based solution
for detecting rival vehicles in the F1Tenth competition environment. The primary
challenge was to accurately identify dynamic objects in real time while adhering to the
computational and mechanical constraints of the vehicle.
The proposed solution integrates advanced image processing techniques with existing
resources such as cameras to generate both RGB and depth images. After evaluating
several algorithms, including SSD, Mask R-CNN, and Reinforcement Learning, the YOLOWorld
model was selected for its balance between accuracy and speed. The vision
module was integrated into the system using the ROS2 platform, with a Docker
environment ensuring stability and isolation.